Lung Nodule Classification in CT Images Using 3D DenseNet

Abstract Lung cancer is the main malignant tumour affecting health of residents in China. Automatically discriminating benign and pulmonary nodules can facilitate early detection lung cancer, which reduces mortality. The rising quantity public available CT datasets made it possible to use deep learning approaches for malignancy classification. Unlike most previous models that focused on 2D convolutional neural nets (CNN), here we explore DenseNet architecture with 3D filters pooling kernels. performance proposed nodule classification was evaluated publicly LUNA16 dataset, a subset image database consortium resource initiative dataset (LIDC/IDRI). It achieved 92.4% accuracy. method provides an independent module encouraging prediction accuracy be easily incorporated computer-aided diagnosis system.